Journal of Molecular Graphics and Modelling 21 (2002) 209–213

Collagen stability, hydration and native state

Inés G. Mogilner, Graciela Ruderman, J. Raúl Grigera

∗

Departamento de Ciencias Biológicas, Instituto de F´ısica de L´ıquidos y Sistemas Biológicos (IFLYSIB) CONICET-UNLP-CIC, 59 No. 789 C.C. 565,

Facultad de Ciencias Exactas UNLP, Universidad Nacional de La Plata, 1900 La Plata, Argentina

Received 22 April 2002; accepted 16 July 2002

Abstract

Molecular dynamics simulations of a collagen-like peptide (Pro-Hyp-Gly)

4

-Pro-Hyp-Ala-(Pro-Hyp-Gly)

5

have been done in order to

study the contribution of the hydration structure on keeping the native structure of collagen. The simulation shows that the absence of
water produces a distortion on the molecular conformation and an increase in the number of intra-molecular hydrogen bonds. This is in
agreement with previous experimental results showing the stiffness of collagen under severe drying and its increase in the thermal stability.
This dehydrated material does not keep, however, the native structure.
© 2002 Elsevier Science Inc. All rights reserved.

Keywords: Hydration; Collagen; Molecular dynamics simulations

1. Introduction

Collagen, the most abundant protein in mammals, has a

particular structural motif, the triple helix, shared only with
some host defense and membrane proteins. Because of its
function as the major structural protein in the extra cellular
matrix and the particularity of the triple helix, it has been ex-
tensively studied

[1,2]

. Also, the model polypeptides chains

that contain the repeating sequence X-Y-Gly, with X and
Y, often proline or hydroxyproline, have been used to in-
vestigate on collagen characteristics. Great tensile strength
and thermal stability are characteristics of collagen fibers,
properties that are grounded on its molecular structure. The
collagen triple helix is formed by three left-handed helices
supercoiled right-handed around a common axis. The col-
lagen protein family has around 20 members; among them
the types I, II, III, V, and XI, found in bones, tendons, skin,
which form periodic fibrils and are well characterized

[3]

.

The function of water as a stabilizing agent of colla-

gen have been often considered and extensive studies on
the hydration properties and its possible influence on the
collagen structure have been done. The existence of water
bridges with potential stabilizing properties has been demon-
strated both for the native collagen

[4–6]

and for different

collagen-like peptides

[7–9]

. However, the concept that wa-

ter is essential to keep the structure has been challenged

[10]

.

∗

Corresponding author. Tel.:

[phone];

fax:

[phone].

E-mail address: [e-mail] (J.R. Grigera).

Thermal stability is easily checked, either in fibers or in

solution

[1,2]

. However, this is not a true test for the proper-

ties of the native state. It is also known that collagen fibers
lose their flexibility upon drying

[12]

. If the dehydration

does not proceed below some threshold value, the flexibil-
ity can be recovered by rewetting. This changes can be cor-
related with the sorption isotherm, in which the threshold
value can be seen as the point at which the resorption curve
follows the same pattern as the desorption one

[11]

. This ir-

reversible change after drastic drying is shared with other,
if not all, proteins

[13,14]

. Reaching the irreversible state

we are faced with a rigid material for which the properties
are no longer those of the native state. Thermal stability, as
measured for instance by the shrinkage temperature (T

s

), is

higher than that for the hydrated (native) state

[15]

. The ex-

istence of a collagen-like substance having thermal stability
in anhydrous environments or having other substitute with
more efficient inductive effect

[10]

cannot be used as a proof

that the water is not essential to maintain the native state.

On the other hand, the static picture of the crystallographic

data may induce to believe that the relatively large amount
of water involved in bridges are rigidly bound to the pro-
tein, having long residence times and low mobility. Under
this assumptions, the entropic cost would be indeed huge.
However, it has been experimentally proved

[4,6]

that water

residence times in the specific sites are in the nanosecond to
sub-nanosecond range. Under the circumstance, it is clear
that the relative high exchange rate of most of the hydration
water must be considered to compute the entropic change.
Moreover, the estimated difference between the chemical

1093-3263/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved.
PII: S 1 0 9 3 - 3 2 6 3 ( 0 2 ) 0 0 1 4 5 - 6

210

I.G. Mogilner et al. / Journal of Molecular Graphics and Modelling 21 (2002) 209–213

potential of the specific hydration water and that of the bulk
is slightly favorable to the bound state

[4]

, ruling out the

entropic argument.

In order to check the effect of drying on colla-

gen at a molecular level, we have performed molec-
ular dynamics simulation of a collagen-like peptide
(Pro-Hyp-Gly)

4

-Pro-Hyp-Ala-(Pro-Hyp-Gly)

5

, that have

been widely used as a model for collagen

[7–10]

. We have

simulated the polypeptide in complete absence of water and
fully hydrated. The results from the simulation support the
idea that dehydration does produce denaturation, although
it increases the thermal stability.

2. Methods

2.1. Computational method

Simulations have been carried out using the GROMOS

package (Biomos n.v.Groningen)

[16]

. The equations of mo-

tion are solved with the leap-frog algorithm, the system was
weakly coupled to a thermal and a hydrostatic bath to work
in the isothermal–isobaric ensemble

[17]

at

T = 300 K and

P = 1.013 × 10

5

Pa.

The time step of integration was held on 0.5 fs. All the

simulation runs were made in Pentium based personal com-
puters running under GNU/Linux. Plots were done either
under MS Windows or in a Silicon Graphics O2 workstation.

The force field of GROMOS was used for collagen in con-

junction with the SPC/E water model

[18]

. In the GROMOS

force field, the interactions between non-bonded atoms are
modeled with a 6–12 Lennard–Jones potential and through
the coulombic electrostatic interactions between the atomic
partial charges.

Fig. 1. Structure of the collagen-like peptide (Pro-Hyp-Gly)

4

-Pro-Hyp-Ala-(Pro-Hyp-Gly)

5

as obtained by X-ray diffractions, by molecular dynamics

simulation in aqueous solution and in vacuo. The three pictures are oriented in the same position to facilitate the comparison (drawn with WebLabViewer).

We have used the SHAKE procedure

[19]

to maintain

rigid bond lengths.

2.2. The system

A series of runs were done with a collagen-like peptide

molecule in vacuo and with 4747 molecules of SPC/E water.
For the hydrated system the average box has dimensions of
3

.853 nm × 3.888 nm × 10.1759 nm. The box size was se-

lected such as to include about five water shells around the
protein and then let to adjust during the simulation at con-
stant pressure. Periodic boundary conditions were applied.

Each system was equilibrated for 50 ps and run for 200 ps.

Averages shown correspond to the last 30 ps. We have used
as a starting point the crystallographic coordinates as given
by Bella et al.

[7]

(Protein data Bank code 1CAG). Six

residues were eliminated at the extremes of the molecule,
including those for which the coordinates were undefined.
With the obvious exception of the simulation in vacuo, we
have included the crystallographic water.

3. Results and discussion

Although the crystal structure is constrained by the econ-

omy of packing of the crystal, we may expect that the molec-
ular structure obtained from crystallography is very close to
the native state. Therefore, we have compared the structures
obtained from the simulations with crystallographic coordi-
nates.

The effect of different conditions on the overall structure

can be observed in

Fig. 1

. A quantitative comparison can be

seen in

Fig. 2

, which shows the root mean square (RMS)

departures of distances of the

␣-carbon atoms when com-

I.G. Mogilner et al. / Journal of Molecular Graphics and Modelling 21 (2002) 209–213

211

Fig. 2. Difference on the average coordinates between the simulation and the crystal structure seen as the RMS of

␣-carbon atoms. Simulations in vacuo

and in aqueous solution.

pared with the crystallographic data for the two simulated
systems. We can see that the results from the simulation in
vacuo show the largest departures from the reference.

We may ask ourselves how do these results compare to

the experimental data. According to the evidence

[1,2]

dehy-

dration produce more rigidity on the fibers and also increase
the thermal stability.

The protein mobility was analyzed from the trajectories

collected along the MD runs.

Fig. 3

shows the RMS for

␣-carbon atoms of each of the simulated systems as an av-
erage over 30 ps. We can see there that the absence of the
solvent make the molecule more stiff, according to the ex-
perimental observations.

Fig. 3. Comparison between the mobility of collagen-like peptide

␣-carbon atoms along the simulations for the in vacuo and in aqueous solution.

We see that around residue numbers 1, 30, 58 and 84

(around atoms 1, 225, 425 and 597, respectively) the mo-
bility, as is seen from the simulation, is higher for all cases.
This corresponds to those residues located around the be-
ginning or the end of the chains, having therefore less con-
strains from the neighboring atoms.

Related to the stiffness of the molecules the most interest-

ing data are the computations of intra-chain hydrogen bonds,
as shown in

Fig. 4

. There the criterion for the existence of

hydrogen bonds was that the angle between donor, proton
and acceptor is larger than 135

◦

and the distance between

the donor and acceptor is equal or shorter than 2.0 nm. We
can see that in the simulation that includes water the sys-

212

I.G. Mogilner et al. / Journal of Molecular Graphics and Modelling 21 (2002) 209–213

Fig. 4. Plot of the hydrogen bonds between residues determined crystallographically for the dry crystal and detected during the simulation of the system
in vacuo and in aqueous solution.

tem presents almost the same number and kinds of hydrogen
bonds than in the crystal, while in the simulation in vacuo
the number of hydrogen bonds increases. Striking are the
large number of bonds near to the diagonal of the graph.
The appearance of this line is considered in other proteins
a signature of the presence of

␣-helix, while the lines per-

pendicular to it represent

␤-sheets

[20]

. The

␣-helix is not

a motif present in native collagen, which has a left-handed
helix. These extra bonds are neither observed in the crystal-
lographic structure nor in those corresponding to the simu-

Fig. 5. Detail of the collagen structure obtained by simulation in which two inner-chain water bridges can be observed, as well of some hydration water.
For the sake of clarity not all the water present in the system are shown (drawn with WebLabViewer).

lation in presence of water. However, for the structures cor-
responding to the simulation in vacuo, several regions are
formed where this bonds are present. Irrespective of that,
it is also seen that the number of hydrogen bonds is larger
for the dehydrated systems, with the consequent increase in
thermal stability; the structure thus stabilized, however, is
non-native.

The existence of water bridges, early suggested by

Ramachandran and Chandrasekran

[21]

and confirmed

by NMR

[4]

, is also seen in the simulations and clearly

I.G. Mogilner et al. / Journal of Molecular Graphics and Modelling 21 (2002) 209–213

213

contribute to the stabilization of the molecules. The more
recent results both by X-ray diffraction

[7]

and by NMR

[6]

shown that water bridges are not only inter-chain but

also intra-chain.

Fig. 5

shows two inter-chain water bridges

(a detail obtained from a snap shot of the configurations
obtained in the MD run). Note that these water bridges may
be connected to the rest of the water network.

It is striking, however, to see the that the conformation

remains quite stable even in the extremes, where the occur-
rence of water bridges is low. This suggests that the overall
hydration around the molecule may contribute through its
network to help on the stabilization process. According to
Bella

[7]

, this hydration shell forms a cylindrical clatharate

like the structure around the collagen. In spite of the fact
that the exchange rate of the water molecules participating
in the hydration shell is relatively high, this quasi clatharate
structure persist

[6]

. As a consequence, the non-specific hy-

dration contribution to the stability of the collagen may not
be only to provide an appropriate environment to keep the
specific hydration molecules (forming the bridges) but also
to help in a direct way.

As we have mentioned before, severe dehydration pro-

duces irreversible changes. Even under rewetting, the rupture
of the extra hydrogen bonds formed is unlikely, producing
an increase in the thermal stability, when compared to the
native state, even if the measurements are done immersed
in water. Notwithstanding that, we are not in presence of a
native state.

4. Conclusions

The molecular dynamics simulation of a collagen-like

peptide shows, in accordance with the experiments, that the
presence of water is essential to keep the native structure
of collagen-like molecules. The absence of water induces
changes in the structure and increases the rigidity of the
molecules. The concept of thermal stability is not by any
means an appropriate test for checking the stabilizing effect
of the media on the native structure.

Acknowledgements

This work was partially supported by the Consejo Na-

cional de Investigaciones Cient´ıficas y Técnicas of Argentina
(CONICET), the Comisión de Investigaciones Cient´ıficas
de la Provincia de Buenos Aires (CIC) and the Universidad
Nacional de La Plata. JRG and GR are members of the Car-
rera del Investigador of CONICET, IGM is granted by the
CONICET.

References

[1] K.H. Gustavson, The Chemistry and Reactivity of Collagen,

Academic Press, New York, 1956, pp. 202–245.

[2] P.H. Von Hippel, Structure and stabilization of the collagen molecule

in solution, in: G.N. Ramachandran (Ed.), Treatise on Collagen,
Academic Press, London, 1967, pp. 253–366.

[3] S. Gay, E.J. Miller, Collagen in the Physiology and Pathology of

Connective Tissue, Gustav Fischer Verlag, Stuttgart, 1978.

[4] J.R. Grigera, H.J.C. Berendsen, The molecular details of collagen

hydration, Biopolymers 18 (1979) 47–57.

[5] E. Suzuki, R.D.B. Fraser, T.P. MacRae, Role of hydroxyproline in

the stabilization of the collagen molecule via water molecules, Int.
J. Biol. Macromol. 2 (1980) 54–56.

[6] G. Melacini, A.J.J. Bonvin, M. Goodman, R. Boelens, R. Kaptein,

Hydration dynamics of the collagen triple helix, J. Mol. Biol. 300
(2000) 1041–1048.

[7] J. Bella, M. Eaton, B. Brodsky, H.H. Berman, Crystal and molecular

structure of a colllagen-like peptide at 1.9 A resolution, Science 266
(1994) 75–81.

[8] J. Bella, B. Brodsky, H.H. Berman, Hydration structure of a collagen

peptide, Structure 3 (1995) 893–906.

[9] R.Z. Kramer, J. Bella, B. Brodsky, H.M. Berman, The crystal and

molecular structure of a collagen-like peptide with a biologically
relevant sequence, J. Mol. Biol. 331 (2001) 131–147.

[10] S.K. Holmgren, K.M. Taylor, L.E. Bresscher, R.T. Raines, Code for

collagen’s stability decipher, Nature 392 (1998) 666–667.

[11] M.

Luscher-Mattli,

Thermodynamic

function

of

biopolymer

hydration. I. Their determination by vapor pressure studies, discussed
in an analysis of the primary hydration process, Biopolymers 21
(1982) 403–418.

[12] M. Luscher-Mattli, Enthalpy–entropy compensation in hydrophilic

hydration processes, Biopolymers 21 (1982) 419–429.

[13] J.R. Grigera, I.G. Mogilner, in: F. Franks (Ed.), Water Bridge in

Myoglobin, Academic Press, London, 1982, pp. 39–41.

[14] A. Salvay, Estudios sobre el rol del Agua de Hidratación en la

estructura y función de la Hemoglobina, Dissertaion, University of
La Plata, 2001.

[15] J.R. Grigera, I.G. Mogilner, Thermal stability of collagen as studied

by length vs. temperature behavior of collagen films, in: Proceedings
of the Congress of International Union of Leather Technologist and
Chemist Societies, Venice, Italy, 15–17 September 1983, pp. 44–49.

[16] W.F. van Gunsteren, H.J.C. Berendsen, GROMOS. From BIOMOS

n.v. Nijenborg 4, 9747 AG Groningen, 1987.

[17] H.J.C. Berendsen, J.P.M. Postma, W.F. van Gunsteren, A. Di Nola,

J.R. Haak, Molecular dynamics coupling to an external bath, J.
Chem. Phys. 81 (1984) 3684–3690.

[18] H.J.C. Berendsen, J.R. Grigera, T.P. Straatsma, The missing term in

the effective pair potentials, J. Phys. Chem. 91 (1987) 6269–6271.

[19] J.P. Ryckaert, G. Ciccotti, H.J.C. Berendsen, Numerical integration

for the Cartesian equations of motion of a system with constrains.
Molecular dynamics of n-alkanes, J. Comp. Phys. 23 (1977) 327–
341.

[20] T. Velazquez, C.T. Reimann, O. Tapia, Proteins in vacuo.

Denaturations of highly charged disulfide-reduced lysozyme studied
by molecular dynamics simulation, J. Phys. Chem. B 104 (2000)
2546–2558.

[21] G.N. Ramachandran, R. Chandrasekharan, Inter-chain hydrogen

bonds via bound water molecules in the collagen triple helix,
Biopolymers 6 (1968) 1648–1658.